This application claims Paris Convention priority of DE 199 54 925.7 filed Nov. 16, 1999 the complete disclosure of which is hereby incorporated by reference.
The present invention is related to the invention xe2x80x9cMethod of Correcting Linear Field Inhomogeneities in a Magnetic Resonance Apparatusxe2x80x9d which was filed claiming prority of DE 199 54 926.5 filed Nov. 16, 1999 the full disclosure of which is hereby incorporated by reference. The two methods can complement one another in an advantageous fashion. The cited method can precede the present method to eliminate linear field inhomogeneities in advance.
The present invention concerns a method of determining and correcting higher order magnetic field inhomogeneities of a nearly homogeneous magnetic field B0 in the investigation volume of a nuclear magnetic resonance apparatus, wherein magnetic resonance is excited in a sample located in the investigation volume through a radio frequency pulse, at least one additional linear magnetic gradient field is applied, and a magnetic resonance signal is measured.
A method of this type is e.g. known from U.S. Pat. No. 5,345,178 for a nuclear magnetic tomography apparatus.
In the known method, a gradient or spin echo sequence is measured after radio frequency excitation of a sample. The NMR signal is then Fourier transformed and a phase profile is determined within a predetermined area. This procedure is repeated for a plurality of projection directions and the phase curves obtained are analyzed with a fit method. Coefficients of a series expansion of the magnetic field dependence in spherical harmonic functions are determined therefrom and correction currents are, in turn, calculated for individual shim or gradient coils associated with the coefficients.
The known method has i.a. the disadvantages that the chemical shift between water and fatty constituents of the signal produces undesired geometric shifts and it cannot be applied or only disadvantageously with samples with short T1/T*2 times.
For this reason, there is a need for a rapid, uncomplicated, direct shimming method which can be made insensitive to the influences of chemical shifts and which can also be successfully applied with samples having short relaxation times.
This object is achieved by a method of the above-mentioned type comprising the following steps:
A) a first radio frequency excitation pulse is irradiated onto the sample;
B) a first phase gradient Gix is applied in a predetermined direction x;
C) at a fixed time tdx1 after the first radio frequency excitation pulse, a value Six1 of the magnetic resonance signal from the sample is measured, digitized and stored;
D) a second radio frequency excitation pulse is irradiated onto the sample;
E) a second phase gradient Gix2 is applied in the predetermined direction x;
F) at a fixed time tdx2 after the second radio frequency excitation pulse, a value Six2 of the magnetic resonance signal from the sample is measured, digitized and stored, wherein Gix2 and tdx2 are selected such that the integrals             ∫      0              tδx        ⁢                  xe2x80x83                ⁢        1              ⁢          xe2x80x83        ⁢          Gx      ⁢              xe2x80x83            ⁢      1      ⁢              ⅆ        t              =            ∫      0              tδx        ⁢                  xe2x80x83                ⁢        2              ⁢          Gx      ⁢              xe2x80x83            ⁢      2      ⁢              xe2x80x83            ⁢              ⅆ        t            
xe2x80x83are identical;
G) the steps A) to F) are repeated several times with systematically altered strength of the phase gradient Gix;
H) the values of the measured resonance signals Six1 are combined, in dependence on the associated gradient strength Gix, into a quasi-spin echo data set Sx1;
I) the values of the measured resonance signals Six2 are combined, in dependence on the associated gradient strength Gix, into a quasi-spin echo data set Sx2;
J) the data set Sx1 is Fourier transformed and optionally phase-corrected such that the phases xcfx86ix1 of the phase-corrected resonance signals Sxe2x80x2ix1 all have essentially the same value;
K) the data set Sx2 is also Fourier transformed and phase-corrected with the same parameters as the data set Sx1 in step J), wherein the difference xcfx86xe2x80x2ix1-xcfx86xe2x80x2ix2 of the phase-corrected phases of the signals Sxe2x80x2ix1 and Sxe2x80x2ix2 represent a measurement for a profile of the magnetic field inhomogeneity along the direction x;
L) in the following measurements of magnetic resonance in the apparatus, a correction magnetic field B(x) is applied in the investigation volume for homogenizing the magnetic field B. which compensates for the magnetic field inhomogeneity determined in step K).
One single measuring point is recorded at each of two defined points in time following each excitation rather than a complete signal echo or FID. The times td1 and td2 are always the same as is therefore the dephasing due to inhomogeneities of the magnetic field B0 . Through application of a phase gradient in the interval between t=0 and t=td1 or between t=0 and t=td2, additional dephasing is produced which can be controlled in a defined manner. At the times of data recording, the two effects overlap. The gradient strengths are selected such that each measuring value at td1 has a corresponding one at td2 which has the same dephasing due to the applied phase gradient but not due to the field inhomogeneity of B0. Evaluation according to the above-mentioned steps produces a phase difference profile in the direction of the applied phase gradient which, except for the T2 relaxation effects, depends only on the effect of the B0 inhomogeneity in this direction. The phase correction mentioned in step J) is optional: it is only important that the same phase correction is applied for both data sets.
The number of required excitations can be reduced if the times td1 and td2 and the gradient strengths are in rational relationship to one another. It is then possible to use one measuring point for at least some gradient strengths of the phase gradient with both td1 and td2, wherein the one at td2 is associated with a measurement at td1 with stronger gradient such that, for this pair, the condition of step F) is met.
The influence of chemical shift is eliminated in that a xe2x80x9cquasi spin echoxe2x80x9d is generated and evaluated in the phase direction rather than in the reading direction, i.e. the measuring points are, in each case, at the same relative point in time (td1, td2) after excitation and do not differ with respect to dephasing through chemical shift. It is recommended to select the times td1, td2 such that the fatty and water contributions of the signals in the B0 field are at least approximately in phase.
The time td can be selected largely freely and can, in particular, be very short for samples with short relaxation times T1/T*2.
The method has been described above with reference to one direction but can be easily applied for several directions. Compensation in two dimensions is advantageous, in particular, for investigations of slices of an object. In multiple slice investigations, the field can be homogenized separately for each slice.
It is of course possible to extend the method in a corresponding manner to three-dimensional volumes by carrying it out analogously for a further dimension to permit homogenization of the field in the entire sample volume or, in connection with volume-selective measures, for selected partial volumes and optionally for many different volumes within an object.
In a particularly advantageous fashion, the method is carried out analogously for a plurality of predetermined directions which are selected such that, according to methods known per se, the principal expansion coefficients of a series expansion of the magnetic field B0 in spherical harmonic functions can be determined, in particular, those of second order. An article by R. Gruetter in Magnetic Resonance in Medicine 29:804-811 (1993) describes e.g. a shimming method with the title xe2x80x9cFastmapxe2x80x9d. In a preferred variant, measurement is carried out along only six projection directions for determining the coefficients up to second order inclusively. These directions can be utilized in the present method in accordance with the invention.
Depending on the shim and gradient coil equipment of the apparatus and depending on the object, the method in accordance with the invention is preferably carried out analogously for a plurality of predetermined directions which are selected such that the correction currents can be determined with methods known per se for shim coils or gradient coils of predetermined correction field profiles. These field profiles may coincidexe2x80x94but must notxe2x80x94with the profiles which are predetermined by the corresponding pure spherical harmonic functions.
This method is preferably a method of nuclear magnetic resonance and, in particular, is incorporated in a method of magnetic resonance imaging. Nuclear magnetic resonance imaging apparatus and, to an increasing extent, high-resolution nuclear magnetic resonance spectroscopy apparatus are provided with gradient coils and shim systems as standard equipment. The method can be integrated into the existing software of, in particular, imaging apparatus without any hardware problems. In particular, it can be added to the actual measurement or be completely integrated therein, such that the field is newly homogenized during the investigation program. As mentioned above, homogenization can be carried out individually for separate partial investigation areas.
In an embodiment of the invention, the method is carried out in two dimensions and is preceded by a slice selection step, possibly including step A), which selects a slice in the investigation object perpendicular to a direction.
Alternatively, it is preceded by a volume selection step, possibly including step A), which selects a volume in the investigation object.
With multiple volume experiments or multiple slice experiments, several sets of correction field profiles are determined which are associated with different volumes or slices.
Correction currents can be determined from the extracted profiles for input into shim coils of the apparatus.
Alternatively, or additionally, one can determine offset currents from the extracted profiles for input into gradient coils of the apparatus thereby allowing particularly simple modification of the currents even during a pulse program. The use of other means for field homogenization instead of shim coils is fundamentally possible, e.g. ferromagnetic or permanent-magnetic shim plates, whose number and/or position must be determined.
In one embodiment, the sample comprises biological tissue. The above-mentioned advantages of the invention have a particularly positive effect with inhomogeneous samples having fatty and water constituents whose exact position might not be known. The same is true for biological samples such as test animals or also human patients. In the inventive method, an inaccurately centered sample produces measurement errors of lower order during determination of the coefficients of a given order which, in turn, can be eliminated through further iterations determining precisely this order.
Further advantages can be extracted from the drawing and the description. The features mentioned above and below can be used in accordance with the invention either individually or collectively in any arbitrary combination. The embodiments shown and described are not to be understood as exhaustive enumeration but rather have exemplary character for describing the invention.
The invention is shown in the drawing and explained in more detail by means of embodiments.